Cerebrovascular Disease

Completness of coding within a group of codes: G45, I60-I69 (3-digit level)
2004 - 2006 QA sample data (extract from SMR01 National data

		Main Condition - code which should have been recorded
		Cerebrovascular disease (CVD)									Non CVD				SMR Total
		G45	I60	I61	I62	I63	I64	I65	I67	total	G81	I72	R55	total
CVD	G45	14						1							15
Main Condition	I60		1												1
as recorded	I61			5											5
on SMR01	I62			1	5										6
	I63					28	1				1		1		31
	I64					5	17				1				23
	I65							8							8
	I66					1									1
	I67								5			1	1		7
QA Total		14	1	6	5	34	18	9	5	92	2	1	2	5	97
Non CVD	G81					2
	I51						1
	M62								1
	R07					1
	R93								1
	S72								1
	T81				1
	Z03							1
QA under-recorded					1	3	1	1	3	9
Number of required codes										101

 

Legend
	Group correctly coded			Group under-recorded
	Group over-recorded			Group recorded on SMR01

 

The Group was recorded with:	Accuracy	95% CI	Sensitivity	Completeness
	94.8%	4.4%	91.1%	96.0%

 

Desciption of ICD10 Codes
Cerebrovascular Disease codes		Non-Cerebrovascular Disease codes
G45	Transient cerebral ischaemic attacks	G81	Hemiplegia
I60	Subarachnoid Haemorrhage	I72	Other aneurysm
I61	Intracerebral haemorrhage disease	I51	Complications and ill-defined description of heart
I62	Other non-traumatic intracranial haemorrhage	M62	Other disorders of muscle
I63	Cerebral infarction	R07	Pain in chest and throat
I64	Stroke, not specified as haemorrhage or infarction	R55	Syncope and collapse
I65	Occlusion and stenosis of precebral arteries, not resulting in cerebral infarction	R93	Abnormal findings on diagnostic imaging of other body structures
		S72	Fracture of femur
I67	Other cerebrovascular diseases	T81	Complications of procedures, nec
		Z03	Medical observation and evaluation for suspected diseases and complications

 

Interpreting the table:
The vertical axis to the left lists the ICD10 codes which had been recorded as the Main Condition in the SMR01 records.  The horizontal axis along the top lists the codes which ISD found should have been recorded.

This table contains 97 records which had a code from the CVD group (G45, I60-I69) recorded as SMR01 Main Condition.

ISD found that 92 of the 97 recorded CVD codes hit the target group, and so we can say that the recorded CVD group codes were (92/97*100)= 94.8% accurate.  In five (5/97*100)= 5.2% of the 97 cases ISD found that CVD should not have been recorded as the main diagnosis - i.e. CVD was over-recorded.

ISD also found that in nine further records CVD should have been recorded as the main diagnosis but was not - i.e. CVD was under-recorded in these records.  As we have already seen, CVD was correctly recorded as Main Condition in 92 records.  This means that in total CVD was or should have been recorded as Main Condition in 101 records.  The total of correct and under-recorded codes can described as the 101 true CVD codes.  This true total can be used to calculate the sensitivity of the recorded group - that is the correctly recorded codes as a percentage of the true total (92/92+9)*100 = 91.1%.

We saw above that CVD was over-recorded in five cases.  Therefore it is possible to calculate a net balance between recorded (correct + over-recorded) and true (correct + under-recorded) codes and express this as 'percentage completeness'.  The CVD group is recorded in Main Condition with ((92+5/92+9)*100)= 96.0% completness - i.e. on balance CVD is 4% under-recorded.